Skip to content

v0.17.0 - Analytical Engine Evolution

Choose a tag to compare

@woconnell1 woconnell1 released this 28 Sep 00:13
· 127 commits to main since this release
e2e3fae

v0.17.0 — Analytical Engine Evolution

v0.17.0 substantially advances the ETIS Engineering Studio's phase-gate reasoning, apprenticeship coaching, evidence analysis, and analytical-control architecture.

Highlights

  • Phase-specific Board Review: A1–A6 executive summaries, prioritized Board Agendas, non-numeric Readiness Maps, repeated-review delta, submission-baseline guidance, and cross-artifact reasoning.
  • Deeper engineering analysis: Reviews reason across claims, evidence, authority, assumptions, contradictions, control maturity, demonstrated outcomes, and claim strength rather than treating artifacts as checklist items.
  • Adaptive apprenticeship coaching: Coaching progresses from challenge/reframe → nudge → scaffold → direct teaching → application/teach-back when needed, while keeping engineering decisions with the student.
  • Progressive-disclosure student UX: The active concern and coaching conversation stay primary, with detailed readiness, evidence, guidance, and findings available on demand.
  • Frozen-evidence artifact review: Students can ask the Board directly about a selected repository artifact and receive semantic feedback grounded in that artifact's frozen review snapshot rather than live repository state.
  • Evidence-safe semantic context: Selected artifact content remains bounded, sanitized, and quarantined where appropriate, with valid structured context maintained under prompt limits.
  • Analytical control plane: Introduces explicit Review Objectives, planning context, candidate next moves, structured selection/rejection semantics, and session-locked analytical modes.
  • Shadow reasoning and planning architecture: Adds independent reasoning validation and planner/selector/realizer infrastructure with fail-closed authority boundaries and internal telemetry.
  • Analytical evaluation and war-gaming: Adds a 42-case A1–A6 engineering reasoning corpus, deterministic replay, replicated acquisition, stability analysis, blinded A/B review infrastructure, and explicit enablement thresholds.

Compatibility and Authority

The analytical shadow capabilities remain non-authoritative unless explicitly enabled for a session. Frozen repository evidence remains authoritative; shadow validation and planning do not independently alter student-visible readiness, recommendations, finding lifecycle, or engineering decisions.

Existing v0.16.1 sessions remain compatible and continue using legacy reasoning and planning when no v0.17 analytical-control state is present.

Validation

v0.17.0 passed the full automated regression suite, CI, shadow deployment, and production-test-student smoke validation before release.

For the complete engineering change history, evaluation calibration, and compatibility details, see CHANGELOG.md and the v0.17 analytical architecture documents in docs/architecture/.